While I realize that any sort of Honeypot based solution such as StopForumSpam and/or Project HoneyPot API keys have a potential for false positives, especially in the cases of Dynamic IP addresses or ISP domain email addresses, the only thing I can do for controlling that is provision them with false positive information whenever possible, which I do.
And I'm not keen on seeing what sort of flood we would be hit with in terms of temporarily suspending either of those functions. While I do realize the abundant frustration caused by getting caught by a "false" positive, I'd rather lean on making it plainly clear that it can happen and how to get in touch with somebody (and what information to provide) so that we can work through each one as they happen.
The other alternative is what we were doing before, and we were still managing to ban-delete or reject valid registrations in a much higher count than that, including accidentally deleting people that had already managed posting genuine content.
That being said, I'm not the "end all be all" for the Community, so if there is a consensus that want's to have a peek at the world with the walls down, I can make it happen.